[evlatests] T304 Attenuators -- gone wild!

[Rick Perley]

    Michael observed a strong source at L-band for about 22 minutes late 
last week.  The script requested about 45 observations, each of (only) 
30 seconds duration.  The integration time was 1 second.  Apparently, 
the 'set and forget' attenuator mode was not enabled, so he requested a 
close look at correlator outputs to see what happened. 

    Three antennas (ea2, ea9, and ea24) showed rapid oscillations in 
correlation output, which can only be explained by having 1 dB 
attenuation switching in and out with a 1 Hz rate.  But the behavior was 
not identical on all three, nor was it the same throughout the short 
observation:

    1) For the first four minutes, these three antennas showed the 
correlation amplitude switching at the 1 Hz rate.  

    2) For the next minute (two scans), the switching stopped, and 
antennas 2 and 9 were stable.  For antenna 24 however, there was a 1 dB 
change in attenuator level between the two 30-second scans. 

    3) For the next 5 minutes, all three antenna resumed the 1 Hz 
switching of the 1 dB attenuator. 

    4) For the remaining 10 minutes of the run, the three antennas 
returned to the mode described in (2) above -- antennas 2 and 9 were 
stable (but with a caveat -- see below), while antenna 24 oscillated 
between two different states, changing every 30 seconds. 

    Caveat:  During the period of 'stability', antennas 2 and 9 showed 
amplitude spikes or dips at the 30-second boundaries separating scans.  
So even through the attenuators apparently did not change between scans, 
something caused a momentary 'glitch' at scan boundaries. 
   
    I wish to emphasize two things:

    1) The three misbehaving antennas were not acting independently -- 
the transitions between one mode of misbehavior and the other were the 
same for all three. 
    2) The misbehavior for antennas 2 and 9 appear exactly identical.  
But antenna 24's misbehavior, although qualitatively similar, is notably 
different in detail.  For those interested, I have plots that will 
display all this much more effectively than any written description. 

    The other 9 antennas were beautifully stable throughout the run, 
with no changes or glitches of any kind. 

    Phase behavior for all 12 antennas was perfect.  The effects noted 
above caused no visible perturbation to phase.